دانلود رایگان مقاله تفسیر تکرار داده ها و بسط

Abstract

Honig and Samuelsson (2014) and Delmar (2015) recently had an exchange in this journal related to a replication-and-extension attempt of two papers which originally arrived at different conclusions based on the same data set. This commentary provides further clarification on the issues and links the debate to broader issues scholarly culture and practices in entrepreneurship research.

Null” findings can be important. The low appreciation of replication is related to the widespread over-reliance on “statistical significance” as a truth criterion (Hubbard and Lindsay, 2013a, 2013b; Schwab et al., 2011; Simmons et al., 2011). Although many individual researchers probably know better, our collective behavior suggests we regard a statistically significant result in a single study (where the assumptions necessary for valid statistical inference were probably already violated) as solid enough evidence to use in the classroom and for policy advice, and to regard subsequent replications as “uninteresting” (usually leading to them not being undertaken). With the focus on statistical significance follows failure to appreciate the importance of absence of expected effects (Landis and Rogelberg, 2013). On the substantive issue the core message in H&K and the current extension is that business planning as such does not lead to better performance of nascent ventures. I applaud H&K's achievement of getting a paper with such a “null finding” as its core message published in a prestigious outlet (and I cannot help wonder whether it would have been possible without the contrast against a “marginally significant” effect on survival to keep reviewers at peace). As the co-author of a recently completed four-year, four-journal journey of the same nature I can testify that convincing colleagues that non-effects can be theoretically and practically important is not an easy task (Davidsson and Gordon, 2015). I also find laudable the current effort to collect and publish additional evidence, stacked against the double institutional norms against “non-significant” results and against replication studies. One might wonder what other important, potentially “mythbusting” null findings allude us because they have not found champions that are willing to stand up against the aberrations of our research culture? Again, we have reason to welcome the arrival of JBVi and its broader appreciation of meaningful scholarly contributions.